Skip to content

Commit

Permalink
add github pages
Browse files Browse the repository at this point in the history
  • Loading branch information
@chen-hao-chao
chen-hao-chao committed Jun 19, 2024
1 parent 0db3b6c commit 63e94c5
Show file tree
Hide file tree
Showing 109 changed files with 3,345 additions and 130,093 deletions.
2 changes: 2 additions & 0 deletions .gitignore
View file
Original file line number Diff line number Diff line change
@@ -0,0 +1,2 @@
.DS_store
.idea
120 changes: 0 additions & 120 deletions README.md
View file
Original file line number Diff line number Diff line change
@@ -1,120 +0,0 @@
# Semantic Segmentation Based Unsupervised Domain Adaptation via Pseudo-Label Fusion

[![MIT License](https://img.shields.io/badge/license-MIT-blue.svg)](LICENSE.md)

### File Structure
```
weights/
├── synthia/
├── gta5/
MMD/
├── train_deeplabv2/
├── train_deeplabv3+/
├── ...
...
Warehouse/
├── SYNTHIA/
│ ├── labels/
│ ├── images/
│ ├── depth/
| | ├── 0000000.png
| | ├── 0000001.png
| | ├── ...
├── GTA5/
│ ├── image/
│ ├── labels/
| | ├── 00000.png
| | ├── 00001.png
| | ├── ...
├── Cityscapes/
│ ├── data/
│ │ ├── gtFine/
│ │ ├── leftImg8bit/
│ │ │ ├── train/
│ | | ├── val/
│ | | ├── test/
│ │ | | ├── aachen
│ │ | | ├── ...
```
### Training
Quick start:
1. Down the pre-generated pseudo label here. (currently unavilable)
2. Place the pseudo label in the `Cityscapes/data/gtFine` folder and train with the following command:
```
cd train_deeplabv3+
python train.py
```

The whole training procedure:
1. Train the teacher models
- [DACS](https://github.com/vikolss/DACS)
- [CRST](https://github.com/yzou2/CRST)
- [CBST](https://github.com/yzou2/CBST)
- [R-MRNet](https://github.com/layumi/Seg-Uncertainty)
2. Generate the pseudo labels and the output tensor
3. Fuse the pseudo labels
```
cd label_fusion
python3 label_fusion.py
```
4. Place the pseudo label in the `Cityscapes/data/gtFine` folder and train with the following command:
```
cd train_deeplabv3+
python train.py
```


### Testing
```
================ GTA5 ================
{ Deeplabv2 }
cd train_deeplabv2
python test.py --restore-from ../../weights/weights/gta5/deeplabv2/resnet/certainty/model_50.13.pth
python test.py --restore-from ../../weights/weights/gta5/deeplabv2/resnet/priority/model_52.96.pth
python test.py --restore-from ../../weights/weights/gta5/deeplabv2/resnet/majority/model_52.76.pth
python test.py --backbone drn --restore-from ../../weights/weights/gta5/deeplabv2/drn/certainty/model_53.83.pth
python test.py --backbone drn --restore-from ../../weights/weights/gta5/deeplabv2/drn/priority/model_54.35.pth
python test.py --backbone drn --restore-from ../../weights/weights/gta5/deeplabv2/drn/majority/model_55.25.pth
python test.py --backbone mobilenet --restore-from ../../weights/weights/gta5/deeplabv2/mobilenet/certainty/model_48.23.pth
python test.py --backbone mobilenet --restore-from ../../weights/weights/gta5/deeplabv2/mobilenet/priority/model_51.35.pth
python test.py --backbone mobilenet --restore-from ../../weights/weights/gta5/deeplabv2/mobilenet/majority/model_50.98.pth
{ Deeplabv3+ }
cd train_deeplabv3+
python test.py --restore-from ../../weights/weights/gta5/deeplabv3+/resnet/certainty/model_51.97.pth
python test.py --restore-from ../../weights/weights/gta5/deeplabv3+/resnet/priority/model_55.12.pth
python test.py --restore-from ../../weights/weights/gta5/deeplabv3+/resnet/majority/model_54.75.pth
python test.py --backbone drn --restore-from ../../weights/weights/gta5/deeplabv3+/drn/certainty/model_54.85.pth
python test.py --backbone drn --restore-from ../../weights/weights/gta5/deeplabv3+/drn/priority/model_57.94.pth
python test.py --backbone drn --restore-from ../../weights/weights/gta5/deeplabv3+/drn/majority/model_57.65.pth
python test.py --backbone mobilenet --restore-from ../../weights/weights/gta5/deeplabv3+/mobilenet/certainty/model_51.64.pth
python test.py --backbone mobilenet --restore-from ../../weights/weights/gta5/deeplabv3+/mobilenet/priority/model_54.74.pth
python test.py --backbone mobilenet --restore-from ../../weights/weights/gta5/deeplabv3+/mobilenet/majority/model_54.95.pth
============== SYNTHIA ===============
{ Deeplabv2 }
cd train_deeplabv2
python test.py --num-classes 16 --source-domain synthia --restore-from ../../weights/weights/synthia/deeplabv2/resnet/model_47.93.pth
{ Deeplabv3+ }
cd train_deeplabv3+
python test.py --num-classes 16 --source-domain synthia --backbone drn --restore-from ../../weights/weights/synthia/deeplabv3+/drn/model_51.76.pth
python test.py --num-classes 16 --source-domain synthia --backbone mobilenet --restore-from ../../weights/weights/synthia/deeplabv3+/mobilenet/model_50.28.pth
```

### Pretrained Weights
You can download the pretrained model here. (currently unavilable)

### Prerequisites
- Python 3.6
- Pytorch 1.5.0

Download the dependencies:
```
pip install requirement.txt
```

### Acknowledgement
The code is heavily borrowed from the following works:
- MRNet: https://github.com/layumi/Seg-Uncertainty
- Deeplabv3+: https://github.com/jfzhang95/pytorch-deeplab-xception
256 changes: 256 additions & 0 deletions index.html
View file
Original file line number Diff line number Diff line change
@@ -0,0 +1,256 @@
<!DOCTYPE html>
<html>
<head>

<meta charset="utf-8">
<meta name="description"
content="Rethinking Ensemble-Distillation for Semantic Segmentation Based Unsupervised Domain Adaptation">
<meta name="keywords" content="Unsupervised Domain Adaptation, Semantic Segmentation, Ensemble-Distillation">
<meta name="viewport" content="width=device-width, initial-scale=1">
<title>Rethinking Ensemble-Distillation for Semantic Segmentation Based Unsupervised Domain Adaptation</title>

<link href="https://fonts.googleapis.com/css?family=Google+Sans|Noto+Sans|Castoro"
rel="stylesheet">

<link rel="stylesheet" href="./static/css/bulma.min.css">
<link rel="stylesheet" href="./static/css/bulma-carousel.min.css">
<link rel="stylesheet" href="./static/css/bulma-slider.min.css">
<link rel="stylesheet" href="./static/css/fontawesome.all.min.css">
<link rel="stylesheet"
href="https://cdn.jsdelivr.net/gh/jpswalsh/academicons@1/css/academicons.min.css">
<link rel="stylesheet" href="./static/css/index.css">
<link rel="icon" href="./static/images/re.png" type="image/x-icon">
<link rel="stylesheet" href="./static/css/custumize.css">
<link rel="stylesheet" href="https://www.w3schools.com/w3css/4/w3.css">

<script src="https://ajax.googleapis.com/ajax/libs/jquery/3.5.1/jquery.min.js"></script>
<script defer src="./static/js/fontawesome.all.min.js"></script>
<script src="./static/js/bulma-carousel.min.js"></script>
<script src="./static/js/bulma-slider.min.js"></script>
<script src="./static/js/index.js"></script>

<script type="text/x-mathjax-config">
MathJax.Hub.Config({tex2jax: {inlineMath: [['$','$'], ['\\(','\\)']]}});
</script>
<script type="text/javascript"
src="//cdn.mathjax.org/mathjax/latest/MathJax.js?config=TeX-AMS-MML_HTMLorMML">
</script>

</head>

<body>

<nav class="navbar" role="navigation" aria-label="main navigation">
<div class="navbar-brand">
<a role="button" class="navbar-burger" aria-label="menu" aria-expanded="false">
<span aria-hidden="true"></span>
<span aria-hidden="true"></span>
<span aria-hidden="true"></span>
</a>
</div>
</nav>


<section class="hero">
<div class="hero-body">
<div class="container is-max-desktop">
<div class="columns is-centered">
<div class="column has-text-centered">
<h1 class="title is-1 publication-title">Rethinking Ensemble-Distillation for Semantic Segmentation Based Unsupervised Domain Adaptation</h1>
<div class="is-size-5 publication-authors">
<span class="author-block">
<a href="https://chen-hao-chao.github.io/">Chen-Hao Chao</a>,
</span>
<span class="author-block">
<a href="https://profiles.stanford.edu/bo-wun-cheng">Bo-Wun Cheng</a>,
</span>
<span class="author-block">
<a href="https://elsalab.ai/about">Chun-Yi Lee</a>,
</span>
</div>

<div class="is-size-5 publication-authors">
<span class="author-block">National Tsing Hua University</span>
</div>

<div class="column has-text-centered">
<div class="publication-links">
<!-- PDF Link. -->
<span class="link-block">
<a href="https://arxiv.org/abs/2104.14203"
class="external-link button is-normal is-rounded is-dark">
<span class="icon">
<i class="ai ai-arxiv"></i>
</span>
<span>arXiv</span>
</a>
</span>
<!-- Journal Link. -->
<span class="link-block">
<a href="https://dl.acm.org/doi/10.1109/TPAMI.2023.3289308"
class="external-link button is-normal is-rounded is-dark">
<span class="icon">
<i class="ai ai-ieee"></i>
</span>
<span>Journal</span>
</a>
</span>
<!-- Video Link. -->
<span class="link-block">
<a href="https://www.youtube.com/watch?v=Ep9VXNt72m4"
class="external-link button is-normal is-rounded is-dark">
<span class="icon">
<i class="fab fa-youtube"></i>
</span>
<span>Video</span>
</a>
</span>
<!-- Code Link. -->
<span class="link-block">
<a href="https://github.com/chen-hao-chao/Rethinking-EnD-SegUDA"
class="external-link button is-normal is-rounded is-dark">
<span class="icon">
<i class="fab fa-github"></i>
</span>
<span>Code</span>
</a>
</span>
</div>
</div>

<div class="column has-text-centered">
&#9654 <b>Keywords: </b>
<!-- <div class="w3-tag w3-round w3-blue" style="padding:3px"> -->
<div class="w3-tag w3-round w3-blue w3-border w3-border-white">
Unsupervised Domain Adaptation
</div>
<!-- </div> -->
<!-- <div class="w3-tag w3-round w3-blue" style="padding:3px"> -->
<div class="w3-tag w3-round w3-blue w3-border w3-border-white">
Semantic Segmentation
</div>
<!-- </div> -->
<!-- <div class="w3-tag w3-round w3-blue" style="padding:3px"> -->
<div class="w3-tag w3-round w3-blue w3-border w3-border-white">
Ensemble-Distillation
</div>
<!-- </div> -->
</br>
<div class="w3-bar w3-border w3-light-grey" style="height:10px; visibility:hidden;"></div>
&#9654 <b>Venue: </b>
<!-- <div class="w3-tag w3-round w3-dark-grey" style="padding:3px"> -->
<div class="w3-tag w3-round w3-dark-grey w3-border w3-border-white">
IEEE/CVF Computer Vision and Pattern Recognition Conference Workshop (CVPRW 2021)
</div>
<!-- </div> -->
<div class="w3-bar w3-border w3-light-grey" style="height:10px; visibility:hidden;"></div>
&#9654 <b>Journal Extension: </b>
<!-- <div class="w3-tag w3-round w3-dark-grey" style="padding:3px"> -->
<div class="w3-tag w3-round w3-dark-grey w3-border w3-border-white">
IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI)
</div>
<!-- </div> -->
</div>
</br>
<div class="w3-bar w3-border w3-light-grey"></div>
</div>

</div>
</div>
</div>
</section>

<section class="section">
<div class="container is-max-desktop">
<!-- Abstract. -->
<div class="columns is-centered has-text-centered">
<div class="column is-four-fifths">
<h2 class="title is-3">Abstract</h2>
<div class="content has-text-justified scroll-wrapper">
Recent researches on unsupervised domain adaptation (UDA) have demonstrated that
end-to-end ensemble learning frameworks serve as a compelling option for UDA tasks.
Nevertheless, these end-to-end ensemble learning methods often lack flexibility as
any modification to the ensemble requires retraining of their frameworks. To address
this problem, we propose a flexible ensemble-distillation framework for performing
semantic segmentation based UDA, allowing any arbitrary composition of the members
in the ensemble while still maintaining its superior performance. To achieve such
flexibility, our framework is designed to be robust against the output inconsistency
and the performance variation of the members within the ensemble. To examine the
effectiveness and the robustness of our method, we perform an extensive set of
experiments on both GTA5 to Cityscapes and SYNTHIA to Cityscapes benchmarks to
quantitatively inspect the improvements achievable by our method. We further
provide detailed analyses to validate that our design choices are practical and
beneficial. The experimental evidence validates that the proposed method indeed
offer superior performance, robustness and flexibility in semantic segmentation
based UDA tasks against contemporary baseline methods.
</div>
</div>
</div>
<!--/ Abstract. -->

<!-- Paper video. -->
<div class="columns is-centered has-text-centered">
<div class="column is-four-fifths">
<h2 class="title is-3">Video</h2>
<div class="publication-video">
<iframe src="https://www.youtube.com/embed/Ep9VXNt72m4"
frameborder="0" allow="autoplay; encrypted-media" allowfullscreen></iframe>
</div>
</div>
</div></br>
<!--/ Paper video. -->
<div class="w3-bar w3-border w3-light-grey"></div>
</div>
</section>

<section class="section">
<div class="container is-max-desktop">
<h2 class="title" style="text-indent: -0.35em;">Poster</h2>
<div class="columns is-centered has-text-centered">
<div class="content has-text-justified scroll-wrapper">
<p class="has-text-centered" style="font-style: italic;"><embed src="./static/images/CVPRW21-poster.pdf" width="980pt" height="550pt"></p>
</div>
</div>
</section>

<section class="section" id="BibTeX">
<div class="container is-max-desktop content">
<h2 class="title">BibTeX</h2>
<pre><code>@InProceedings{Chao_2021_CVPR,
author = {Chao, Chen-Hao and Cheng, Bo-Wun and Lee, Chun-Yi},
title = {Rethinking Ensemble-Distillation for Semantic Segmentation Based Unsupervised Domain Adaption},
booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops},
month = {June},
year = {2021},
pages = {2610-2620}
}</code></pre>
</div>
</section>


<footer class="footer">
<div class="container">
<div class="columns is-centered">
<div class="column is-8">
<div class="content">
<p>
The template of this page is based on the
<a href="https://github.com/nerfies/nerfies.github.io">Nerfies</a> website.
You are free to borrow the <a href="https://github.com/chen-hao-chao/dlsm">code</a> of this website,
we just ask that you link back to this page in the footer.
Please remember to remove the analytics code included in the header of the website which
you do not want on your website.
</p>
<p>
This website is licensed under a <a rel="license"
href="http://creativecommons.org/licenses/by-sa/4.0/">Creative
Commons Attribution-ShareAlike 4.0 International License</a>.
</p>
</div>
</div>
</div>
</div>
</footer>

</body>
</html>
Loading

0 comments on commit 63e94c5

Please sign in to comment.